Post subject: Help with Smith Chart for Matching at
433MHz Posted: Sat May 16, 2009 3:58 pm
Sat May 16, 2009 3:16 pm
I'm new to Smith
Charts, I've done a lot of reading and understand the concepts but am
having a little difficulty navigating my first problem.
to design a circuit to match a BAT54 diode to a 50ohm source for the
purpose of creating a low-cost, ASK detector at 434MHz. Hoping to achieve
useful sensitivity down to about -50dBm.
I've forward biased
the BAT54 with 10uA and measured the S11 using my network analyser.
It looks mostly open circuit and slightly inductive, see attached plot
which represents 400 to 450MHz.
Is there anyone out there who
can help me? If someone's got the time, I'd prefer to get my head around
the approach and method rather than just be given the solution.
Post subject: Re: Help with Smith Chart for
Matching at 433MHzPosted: Sun May 17, 2009 2:33 am
Joined: Mon Jun 27, 2005 2:02 pm
Is it possible in your application to increase
the forward bias to 100uA? It should give you an easier starting matching
Regarding matching, if the diode is meant to work as
detector, use only reactive components (L,C) for matching network and
not any resisitive attenuators. If you wish to find the network yourself
and optimize it, there are many free softwares in the Internet for impedance
matching, in which you can define the type of network, frequency range,
input and output impedances, Q etc.
Re: Help with Smith Chart for Matching at 433MHzPosted: Sun May 17,
2009 3:07 am
Joined: Sat May 16, 2009 3:16 pm
Increasing bias is no problem, I picked
10uA because it seemed to give the best sensitivity around there, but
I'm sure I'm loosing most the power due to lack of matching anyway.
I understand R's will add loss.
Could you recommend a free
CAD prog for this type of problem?
But if possible, this time,
I'd prefer to do it on paper to help my understanding. Or is this making
my life unnecessarily complex?
My biggest problem is knowing
what type of network to build. Can you give me any guidance here? Although
I've been designing electronics for 20+ yrs, I am very new to this stuff
and I was hoping someone could walk me through the solution so I can
learn then approach.
Post subject: Re: Help with Smith Chart for Matching at 433MHzPosted:
Sun May 17, 2009 3:36 am
Joined: Mon Jun
27, 2005 2:02 pm
Yes indeed, most of the power is lost due to power mismatch.
At first search, I found this SW - http://www.zzmatch.com/ you can
download it for 20 days free.
Doing the calculation on paper
is possible although extremely inconvenient and prone to errors. I don't
think it will bring much to your understading.
network topology, from the given S11 (At 10uA), it looks like the network
would have at least 3 elements: series L, shunt C and another series
L, but of course there are other cominations which would yield 50 ohm
Post subject: Re: Help with
Smith Chart for Matching at 433MHzPosted: Mon May 18, 2009 12:57 am
Joined: Tue Feb 27, 2007 6:16 am
http://bwrc.eecs.berkeley.edu/Research/ ... Match.html
Regards, Darcy Randall, Perth, Western Australia
Post subject: Re: Help with Smith
Chart for Matching at 433MHzPosted: Tue May 19, 2009 1:54 am
Joined: Sat May 16, 2009 3:16 pm
Thanks guys, it's been very helpful. I have downloaded ZZmatch and Microwave
Office, which are fantastic tools but I am still struggling.
The diode looks more or less like an open circuit even with 100uA or
more of bias.
The obvious answer seems to be a length of transmission
line but I don't have space for it at 434MHz.
Is it possible
to transform to an open circuit with lumped elements whilst avoiding
R's? Everything I try just seems to move the match point around the
outside edge of the smith chart, which is obviously the last place I
want to be!
I'm assuming the BAT54 diode looks like 1-2pF's but
the Network Analyser shows the unmatched S11 to be slightly inductive
(below the centre) but its so slight that this measurement could be
being influenced by my test board etc. It's really so close to an open-circuit
that I'm assuming it is, but if that's true, how can any power ever
couple into it. HELP, what am I doing wrong?!?
I only need a
very narrow bandwidth at 434.0MHz, which I'm sure must help.
This is my first attempt at building a matching network, so I've got
no experience on my side and it looks more challenging than any of the
text book examples I've researched.
Many thanks if anyone can
guide me through this.
Re: Help with Smith Chart for Matching at 433MHzPosted: Wed May 20,
2009 9:41 am
Joined: Fri Feb 17, 2006 12:07
Location: London UK
I have been lurking
on this topic since you posted it, because I am building a simple (I
hope) diode power meter, and also need to optimize its sensitivity.
This inevitably involves matching optimally to the diode.
sheet for the BAT54 states the capacitance is around 10pf but shows
it critically depends on bias.
Looking at the Smith Chart, it looks
like you have an impedance of around 300 ohms in parallel with about
3pf capacitance. To match this roughly to 50 ohms, some formulae I have
seen suggest a simple PI matching circuit might get you in the ball-park.
At 434MHz you would need to model into any software you find, an input
(ie 50 ohm port) shunt capacitor of around 8pf, a series L of about
200nH (quite feasible at 434MHz) and a shunt capacitor at the 300 ohm
end of about what the diode is showing, 2 to 3pf. To improve the match
without really decreasing the sensitivity too much, I would terminate
the assembly at the diode end by a shunt 390 ohm chip resistor.
whole subject of matching to detector diodes under widely varying drive
levels occupied the best non-linear theory gurus (like Prof D.G.Tucker)
from the start of World war II right through to the present. It is a
knotty problem that does not seem to have a simple circuit to do it,
so far as I can see. But over a narrow band you might find that matching
technique is good enough.
At bottom, life
is all about
Sucking in and blowing out.
Post subject: Re: Help with Smith Chart for Matching
at 433MHzPosted: Fri May 22, 2009 2:16 am
Sat May 16, 2009 3:16 pm
Thanks for all your help guys!
I think I'm there now.
I think my initial measurements were wrong
and the smith chart I've posted should be ignored, although I'm not
sure how I managed that!
Thanks 'nubbage' for the big clue that
the BAT54's datasheet says it's C is around 10pF, this made me realise
my readings and assumptions must be wrong and the diode shouldn't be
looking like an open circuit. I went back to the beginning, re-cal'ed
everything and started again. Sure enough it looked totally different;
I actually measure around 21pF.
I calculated that a shunt inductor
of 6.4nH should bring it to resonance. With a little experimentation
I found 3.3nH actually created a nice tank circuit at 434MHz and gives
me a nice 50ohm resistive impedance. By adjusting the bias from 10 to
40uA I can fine-tune its resonance to end up with a fairly decent S11
match of -24dB.
I never did manage to model it successfully,
but I think its probably my ignorance with Microwave Office that's causing
the problem, not the design. This looks like an amazing package but
will obviously take some time to master!
I'm now getting usable
DC sensitivity of >5mV with -40dBm input, which is a 25dB improvement
on where I started, so I'm very happy!
Once again, thanks for
all your help